Closed
Bug 1841346
(webgpu-webllm)
Opened 2 years ago
Closed 5 months ago
Ensure that WebLLM works well under WebGPU
Categories
(Core :: Graphics: WebGPU, task, P3)
Core
Graphics: WebGPU
Tracking
()
RESOLVED
FIXED
People
(Reporter: ErichDonGubler, Unassigned)
References
(Blocks 2 open bugs, )
Details
While open, Bug 1636992 noted some performance issues that were discovered while attempting to use WebGPU in Firefox for machine learning. We do not have any evidence of performance problems now, but we'd like to follow up with evaluating WebGPU's performance in machine learning test cases. :tqchen helpfully recommended that we use https://mlc.ai/web-llm/.
Action item: Evaluate the performance of https://mlc.ai/web-llm/ on Firefox once webgpu-v1
EDIT: s/webgpu-v1/bug 1956070
|
Reporter | |
Updated•2 years ago
|
|
|
Reporter | |
Updated•2 years ago
|
Alias: webgpu-v1-webllm
|
|
Reporter | |
Updated•2 years ago
|
Alias: webgpu-v1-webllm → webgpu-webllm
|
||
Updated•1 year ago
|
Blocks: webgpu-triage
|
|
Reporter | |
Updated•5 months ago
|
Depends on: webgpu-f16
|
|
Reporter | |
Comment 2•5 months ago
|
||
Validated that an f16-using model (Llama-2-7b-chat-hf-q4f16_1-MLC) on WebLLM works just fine. Woot! 🙌🏻 Closing.
Status: NEW → RESOLVED
Closed: 5 months ago
Resolution: --- → FIXED
You need to log in
before you can comment on or make changes to this bug.
Description
•